Tips for Simplifying LLM Writing Prompts

Building with AI = Failing A Lot Take writing custom GPT instructions. I've probably written 50+ at this point and for the most part they have all followed a similar markdown pattern I learned over a year ago from Rachel Woods. But models evolve and what worked yesterday might not work as well today. When GPT-4o came out a few weeks ago, I definitely noticed some of my custom GPTs acting weird. Some got better. Some got worse. Some straight up broke. So I went back to the lab again and started to rebuild. Lots of failures later, I now have a new custom GPT prompt structure that still has the bones of my older markdown method, but incorporates a lot of OpenAI's recent guidelines. And now I have GPTs performing better than ever. You can check out the full article, but here are some good guidelines for any prompt writing: ✔ Simplify Complex Instructions -break larger steps down ✔Structure for clarity - use delimiters and examples ✔Promote Attention to Detail - encourage the model to focus on certain areas of the prompt ✔Avoid Negative Instructions - frame instructions positively ✔Granular Steps - break down steps as granularly as possible ✔Consistency and Clarity - be explicit with terms and be sure to define what you want with examples.

𝗛𝗼𝘄 𝗳𝗿𝗲𝗾𝘂𝗲𝗻𝘁𝗹𝘆 𝗱𝗼 𝘆𝗼𝘂 𝗴𝗲𝘁 𝘁𝗵𝗲 𝗿𝗶𝗴𝗵𝘁 𝗼𝘂𝘁𝗽𝘂𝘁 𝗳𝗿𝗼𝗺 𝗖𝗵𝗮𝘁𝗚𝗣𝗧 𝗼𝗻 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝘁𝗿𝘆? 𝗡𝗼𝘁 𝗲𝗻𝗼𝘂𝗴𝗵? Give these tips a try, check out the attached anatomy of a prompt below. Be Specific: Precision in prompts leads to targeted and useful AI responses. It’s about asking the right questions to get the right answers. Set Constraints: Constraints guide the AI in generating focused and relevant outputs. Think of them as guardrails that keep the AI on track. Provide Context: Context is king. It helps AI understand the 'why' behind a prompt, leading to more meaningful and insightful responses. Seek Creativity: Don't shy away from asking for imaginative or out-of-the-box ideas. AI can surprise us with its creative capabilities. Use Clear Language: Clarity is critical. Clear prompts result in clear responses. Avoid ambiguities to ensure that AI understands your exact needs. Include Criteria for Success: Define what success looks like for your prompt. This helps in evaluating the AI's response and in iterative improvements. Ask for Reasoning: Encourage AI to not just provide answers, but also the rationale behind them. This deepens understanding and trust in AI outputs. Iterate and Refine: AI prompting is an iterative process. Refine your prompts based on responses to achieve the best outcomes. 𝗭𝗲𝗿𝗼, 𝗦𝗶𝗻𝗴𝗹𝗲, 𝗮𝗻𝗱 𝗙𝗲𝘄 𝗦𝗵𝗼𝘁 𝗣𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴 In addition to the core strategies for crafting AI prompts, understanding the nuances of zero-shot, single-shot and few-shot prompting can improve your ChatGPT results. 𝗦𝗶𝗻𝗴𝗹𝗲-𝗦𝗵𝗼𝘁 𝗣𝗿𝗼𝗺𝗽𝘁𝘀: 𝗧𝗵𝗲 𝗔𝗿𝘁 𝗼𝗳 𝗣𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻 Typically we don't provide any examples to our prompts this is called zero-shot prompting. Single-shot prompting involves providing the AI with one example. This is ideal for straightforward tasks or when you need a quick, creative solution without much context. The key here is specificity and clarity. Since you're only giving one shot or example it's typically good for showing the format of the output you are looking for. 𝗙𝗲𝘄-𝗦𝗵𝗼𝘁 𝗣𝗿𝗼𝗺𝗽𝘁𝘀: 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝘄𝗶𝘁𝗵 𝗘𝘅𝗮𝗺𝗽𝗹𝗲𝘀 Few-shot prompting means providing the AI with a small number of examples to guide its output.Perfect for tasks where you want the AI to follow a certain style or format, or when more complex understanding is required.Choose your examples wisely. They should be representative of the task at hand and demonstrate the variety you expect in responses. Mastering these skills can significantly enhance the results from ChatGPT and other chatbots that use LLMs. What tips do you have for crafting effective AI prompts?

7 Smart Ways To Write Better AI Prompts (Stop blaming the AI. Start improving your prompts.) Most people write prompts like they’re texting a friend. Then wonder why the results feel off. Here’s how to fix that: 1. Be Specific: ↳ Vague input = vague output. Say exactly what you want. 2. Add Context: ↳ Tell the AI who it is and what it’s doing. Example: “You’re a marketing expert writing a LinkedIn post.” 3. Use Constraints: ↳ Set clear rules like “write under 1000 characters” or “use 3 bullet points.” 4. Test Variations: ↳ Try 3 different prompt styles to compare results. 5. Chain It: ↳ Break big requests into smaller steps and feed the AI output back into itself. 6. Role Play: ↳ Assign the AI a role. It focuses the tone and response. 7. Refine & Iterate: ↳ Great prompts are rarely perfect on the first try. Edit and improve. AI is only as smart as the questions you ask. Are you writing your prompts with intention and structure? ___________________________ AI Consultant, Course Creator & Keynote Speaker Follow Ashley Gross for more about AI

Take small steps when using a LLM. One common mistake is when people give a LLM a single prompt for what is actually a multi step process. A great example is people who ask ChatGPT or Claude to write them a blog on a certain topic. The LLM will do it but not the greatest job. Instead break it down into steps. First ask it to give you a list of catchy titles based on a topic. Then choose one of those titles. After that ask it to write an outline for the blog. Next ask it to research for the outline of the blog. Then ask it to write a draft of the blog based on the topic and research. And lastly ask it to edit the blog for grammar, spelling and clarity. In each step you also want to ask it to play the role of what type of person would perform each task, ie a writer, an editor, a researcher, etc. You will find you end up with a much better product. Happy Sunday all.